Vehicle Dynamic Modelling and Parameter Identification for an Autonomous Vehicle

Abstract

For autonomous vehicles to be feasible, a fast and accurate model of the vehicle dynamics is required due to the complexity of the task. There are many different aspects to a driverless vehicle, including path planning, image processing, data analysis, and the low level control of the vehicle. All these processes are important; they need to work in tandem for the vehicle to be able to drive itself. Regardless of how good all the components are, the vehicle itself must be able to follow the desired trajectory. This is accomplished through the low level control of the vehicle, by using an accurate vehicle dynamic model to assess the safety and feasibility of a given trajectory. This thesis develops a 14 degree of freedom full car model of a 2015 Lincoln MKZ hybrid vehicle. A vehicle measurement system is attached to the vehicle in order to measure the suspension displacement along with the tire orientation, velocities, forces, and moments. In addition, a GPS and an inertial measurement unit is used to measure the position, acceleration, and angular velocities of the chassis. The vehicle is then tested on a dedicated test track in order to identify the vehicle parameters. The center of mass, wheel and vehicle inertias, coefficient of drag, and suspension parameters are identified. In addition, combined slip Pacejka tire models are developed. These parameters are identified using a two-step process. Parameters are first identified using simple physics based models. The second step uses the full vehicle dynamic model to further optimize the parameters, accounting for the numerous simplifications assumed in the simple physics based models. The vehicle dynamic model is implemented and validated in MapleSim 2017.3. The model is intended to be used for controller development and autonomous vehicle testing in a simulation environment.